Viva questions Flashcards
Why PCA (and not eg Factor Analysis)?
PCA and factor analysis are similar in many ways and often present very similar results and are both data reduction techniques.
PCA aims to create one or more components from a larger set of measures through linear combination of these variables. Instead, FA is a model of the measurement of a latent variable, which cannot be measured by a single variable and is measured through the relationships obtained between a set of variables.
Budaev suggests that PCA is generally more appropriate when the objective is to reduce the number of dimensions while FA is more appropriate when the aim is to determine and assess unobservable behavioural constructs.
In addition, FA generally requires a very large sample size which was not possible to obtain in this case.
Furthermore, the vast vast majority of personality assessments use PCA which makes comparison of results easier.
Why did you not perform Varimax rotations?
Essentially varimax is only needed for interpretation of the results; however, due to the small sample size when varimax rotation was performed some PCs did not have any heavily loaded variables; therefore it was considered best to continue without rotation.
What would you change of your methods if you could re-do the study?
In an ideal scenario I would:
- Use a bigger sample size
- Conduct behavioural observations and NOTs in different times of the year in order to test for consistency across context
- Use two novel objects in situations in which pandas are housed together
Why pearsons for keeper survey and spearman for correlations?
I decided to do pearson’s on the survey data because I wanted to test for linear relationships between the traits in order to obtain personality dimensions of significantly correlated behaviours.
Instead, when correlating the results from the 3 methods I decided to use Spearman because I wanted to test for monotonic relationships.
Tell me about your project in 1 minute
So I conducted a personality assessment in red pandas as personality of the mother has been suggested to impact maternal care and cub survival of the species, but little information on personality is available. I assessed personality through a combination of the three most common methods of doing so, involving both rating and coding methods in the form of behavioural observations, NOTs, and keeper surveys. All methods used were able to create personality dimensions which were used to attribute a score to the individual panda. Correlation analyses between the three methods found several significant correlations which were logical and in the predicted direction suggesting that the three methods assessed the same constructs.
What method would you recommend zoos use?
The choice of which method should be used by zoos to assess personality will depend on the aim of the assessment. In general, while behavioural observations are time-consuming, and keeper survey responses rely on experienced keepers, NOTs represent a quick and efficient way to assess personality. However, ideally, a combination of both rating and coding methods would be used in order to increase the validity of results.
Were 5-minute intervals between observations enough?
This was something that I was undecided on; however, in the end it seemed like the 5-minute interval was enough of a wash-out period between one beahviour and the other. In fact, other than when individuals where sleeping, it seemed like they would rarely perform a behaviour for more than 5 minutes.
Implications of carrying out NOTs on the same day – was the pause between the two trials enough?
Carrying out NOTs on the same day was not ideal and in an ideal scenario I would have carried them out with several weeks or months between the two trials; however, due to the budget and time frame I couldn’t spend more than the allocated time at each zoo and therefore had to conduct both trials on the same day. This may have for example reduced curiosity for the second item, taking longer to approach it, or spending less time with the object.
Why did you choose this topic? Why is it important?
Personality has been shown to influence many aspects of captive management including reproduction, welfare, and health and for the red panda, in particular, captive breeding is notoriously challenging and many factors appear to be associated with reproductive success in captivity including stress levels, number and location of nest boxes, vertical space, and enclosure size and quality (Eriksson et al., 2010; Loeffler, 2011). However, data also show that there is a significant difference in breeding success between different zoos, and climate is suggested as an explanation (Princée & Glatston, 2016). In fact, evidence suggests that, as external temperatures increase, mothers spend substantially less time in the nest box with their cubs, leading to the suspicion that higher infant mortality in regions with higher temperatures is due to the fact that maternal behaviour might be compromised in these conditions. And with the recommendation of personality. Also, AZA recommends personality of the mother be considered in the management of cubs and dams; however very little information on personality is available. For these reasons, and my personal appreciation and fascination for the species, I decided to conduct my research project on this topic.
What practical applications do your results have?
Essentially, this study contributes to the understanding of red panda personality, which is unfortunately understudied. The results of this study suggest that the three methods assess the same constructs, suggesting that personality can be rapidly assessed in a captive environment the use of novel object trials, for example. This can inform captive management of individuals, although further research is undoubtedly needed due to the small sample size and additional research is needed in order to assess the impact of red panda personality on maternal care and infant mortality to potentially improve the quality of life of this species.
Implications of having NOTs at the same time
This likely impacted the results of the individuals; however separating them was considered to be more impactful due to the stress that this would have provoked.
Why did you use both PCA results and behaviours/variables? What was the point of doing the PCA then?
Results from PCAs, on NOTs in particular, did not produce much more information than the variables alone and showed that almost all variables loaded on the PCs and those that did not very nearly did. For this reason, it was considered appropriate to test correlations using both PCA results and individual variables.
What were your main findings?
The main findings of this study is that results from behavioural observations, NOTs, and keeper surveys were reliably correlated, suggesting that these methods assess the same constructs in the species. Furthermore, this study highlights behavioural dimensions found in the red panda and opens up to further research in this area.
How similar are your findings to other research?
There is only one other peer-reviewed study on red panda personality which was based on PCA on behacioural observations in 8 individuals. Two PCs were identified which were labelled Maintenance and Exploratory. The exploratory dimension is comparable to our exploratory dimensions while the maintenance dimension is comparable to the Vigilant dimensions found in our study. Other studies also found correlations between different methods and in particular, our correlation coefficient was similar to that found by Gartner & Powell and higher than that found my several other studies.
Biological explanation for why neophobic was also curious
May simply be due to the small sample size and the fact that 2 out of the 6 pandas did not interact with either novel object, 3 only interacted with one and only 1 interacted with both.
